Knowledge distillation for end-to-end speech recognition based on Conformer model

Jiangkun Sang; Yolwas Nurmemet

doi:10.1117/12.2659711

23 November 2022 Knowledge distillation for end-to-end speech recognition based on Conformer model

Jiangkun Sang, Yolwas Nurmemet

Proceedings Volume 12454, International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022); 124542I (2022) https://doi.org/10.1117/12.2659711
Event: International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022), 2022, Hohhot, China

Abstract

With the rise of deep learning, the end-to-end speech recognition model has received increasing attention. Currently, the performance of the end-to-end speech recognition model has been further updated on basis of on the proposal of the Conformer Framework, which has been widely used in the field of speech recognition. However, these models perform poorly on edge hardware due to large memory and computation requirements. This paper compresses the model by exploring the application of knowledge distillation in end-to-end speech recognition based on Conformer. By combining with the network structure of the model, the training loss function is redefined, and on this basis, the influence of different temperatures on the performance of the student model in knowledge distillation is explored.

Citation Download Citation

Jiangkun Sang and Yolwas Nurmemet "Knowledge distillation for end-to-end speech recognition based on Conformer model", Proc. SPIE 12454, International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022), 124542I (23 November 2022); https://doi.org/10.1117/12.2659711

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available